Devanshu
§ PLATE 01·CHILD DEVELOPMENT SCREENING PWA·COMMUNITY ENGAGEMENT

SPARSH

Smart Platform for Assessment of Response, Screening & Holistic Development

Mobile-friendly Progressive Web App that helps Anganwadi workers screen children aged 0–6 for possible developmental concerns using age-based milestone questions, with screening history for follow-up.

Project RoleTeam Leader · 11-member Team
Sem III Community Engagement
Institutional ContextCSE (AI & ML)
PRPCEM Amravati
++++
PLATE 01.1 · ARCHITECTURAL SCHEMATICANGドANWADI COHORT SPECIFICATION
DEVELOPMENTAL MILESTONE TAXONOMY

SPARSH Core Screening Framework

Target Demographic: Children Aged 0–6 Years · 5 Clinical Observation Domains

01
Gross Motor
Balance, posture, locomotion
02
Fine Motor
Object grasp, manipulation
03
Language
Vocal, receptive, expressive
04
Cognitive
Reasoning, exploration, recall
05
Social-Emo
Interaction, affect, response
TIER: REACT 19 PWA + FASTAPI + CLOUD FIRESTOREOUTCOMES: LOW CONCERN · MONITOR · FURTHER EVAL
FIGURE 01.1 · TYPOGRAPHIC FIELD SPECIFICATIONMOBILE-OPTIMIZED PROGRESSIVE WEB APP
§ 01·PROBLEM STATEMENT

The Surveillance Gap in Early Childhood Care

In grassroots public healthcare setups across India, Anganwadi centres serve as the vital first line of observation for millions of children. However, routine monitoring is almost exclusively oriented around physical anthropometric metrics — specifically tracking weight and height.

Critical early delays in physical movement, expressive speech, receptive hearing, and cognitive thinking often pass unnoticed during the foundational 0–6 developmental window. Furthermore, existing state applications such as Poshan Tracker are engineered around supplementary nutrition and ration accounting rather than structured developmental milestone screening.

§ 02·FIELD WORK & PRIMARY INVESTIGATION

Direct Field Visit and Pediatric Validation

To ensure our screening methodology rested on sound pediatric science rather than abstract assumptions, our team initiated a multi-stage investigation:

  • Literature Review: Grounded the milestone catalog in established international and national pediatric guidelines, analyzing standards published by WHO, CDC, UNICEF, the Indian Academy of Pediatrics (IAP), and the Rashtriya Bal Swasthya Karyakram (RBSK).
  • Field Visit (5 October 2026): Conducted an on-ground investigative visit to Anganwadi centres situated in Revasa, Amravati to observe daily operational workflows, record infrastructure constraints, and assess worker digital readiness.
  • Structured Interviews: Completed structured interviews with 4 active Anganwadi workers on-site. All 4 workers confirmed that an intuitive, phone-based screening assistant would be readily integrated into their routine community rounds.
  • Pediatrician Validation: Reviewed milestone question formulations, screening age brackets, and referral thresholds directly with practicing pediatricians to safeguard clinical appropriateness.
§ 03·FUNCTIONALITY

Screening Workflow & Assessment Engine

§ 03.1

Child Registration & Age Engine

Registers each child with date of birth and automatically computes exact chronological age into clinical developmental brackets.

§ 03.2

5 Developmental Domains

Screens across gross motor, fine motor, language, cognitive, and social-emotional milestones tailored to the child's exact bracket.

§ 03.3

Standardized Responses

Questions feature structured, low-friction answers: Yes, No, or Unsure, designed specifically for rapid assessment in field environments.

§ 03.4

Domain-Wise Outcomes

Calculates domain-level outcomes categorized as Low Concern, Monitor, or Further Evaluation to flag potential delays.

§ 03.5

Screening History for Follow-Up

Logs completed assessments over time so Anganwadi workers can track longitudinal trajectory across repeat community visits.

§ 03.6

Mobile-First PWA Design

Installable directly to home screens from the browser with lightweight resource usage suitable for basic field smartphones.

§ 04·SCREENING ENGINE

Rule-Based Determinism & Development Pipeline

SPARSH is currently powered by a transparent, fully deterministic rule-based screening engine. Responses across the 5 developmental domains are evaluated against validated age-specific milestone thresholds, providing immediate, predictable outcome classifications without opaque black-box processing.

ENGINE STATUS NOTICE:The live production deployment relies strictly on the validated rule-based engine. An advanced AI-powered risk pattern engine is currently IN DEVELOPMENT to assist with longitudinal variance detection. No AI model is live in production today, and no artificial accuracy benchmarks are claimed.
§ 05·SYSTEM ARCHITECTURE

Three-Tier Technical Blueprint

Tier 01 · Frontend PWAReact 19 + Vite + Tailwind CSS

Progressive Web App shell delivering rapid response times, touch-friendly questionnaire interfaces, and field installation.

Client Layer
Tier 02 · Service APIFastAPI + Pydantic Engine

Python backend handling strict schema validation, chronological age derivation, and milestone threshold scoring.

Compute Engine
Tier 03 · Data & AuthFirebase Authentication + Cloud Firestore

Role-based Anganwadi worker authentication, secure child record segregation, and historical screening log persistence.

Persistence Layer
§ 06·SYSTEM BOUNDS & LIMITS

Operational Boundaries and Ethical Constraints

Building responsible healthcare technology requires transparency about what an engineered system cannot and must not claim to do:

  • Screening Support Only, Not a Diagnosis: SPARSH is exclusively a preliminary developmental surveillance aid. It does not provide medical diagnoses or replace licensed clinical evaluations by pediatric specialists.
  • No Pilot Deployment Yet: While primary field interviews with 4 Anganwadi workers in Revasa and pediatric consultations have been successfully conducted, an official administrative pilot across public primary health infrastructure has not yet launched.
  • AI Engine in Development: Automated machine learning models for anomaly detection and developmental clustering remain in active research and prototyping; they are deliberately withheld from live screening workflows until comprehensive clinical validation.

Technologies Used

React 19ViteTailwind CSSPWAFastAPIPydanticFirebase AuthCloud Firestore